A5002647742- Machine Learning and Algorithms
- Sparse and Compressive Sensing Techniques
- Stochastic Gradient Optimization Techniques
- Tensor decomposition and applications
- Machine Learning and ELM
- Face and Expression Recognition
- Fetal and Pediatric Neurological Disorders
- DNA and Biological Computing
- Neonatal and fetal brain pathology
- Congenital gastrointestinal and neural anomalies
- Computability, Logic, AI Algorithms
- Blind Source Separation Techniques
- Domain Adaptation and Few-Shot Learning
Nanjing Medical University
2021
University of Kansas
2016
In order to realize the automatic recognition and diagnosis in ultrasound images of fetal spina bifida, U-Net algorithm was improved this study obtain a new convolutional neural network algorithm—Oct-U-Net. 3,300 pregnant women were selected as research objects, who underwent three-dimensional (3D) examinations. Then, Oct-U-Net applied evaluate diagnostic effect bifida by recall rate, precise mean standard error, pixel accuracy (PA), intersection over union (MIoU), running time. Besides,...
Active matrix completion (AMC) is an effective approach to improve the performance of completion. It actively acquires certain missing entries a target matrix, with aim quickly improving accuracy rest. Although this topic attracting increasing attention, all existing solutions are heuristic. In paper, we propose new active called Factor-Disagreed AMC (FDAMC), provably guarantee. extension popular disagreement-based statistical learning task, at both methodological and theoretical levels....
Many classification problems involve instances that are unlabeled, multi-view and multi-class. However, few technique has been benchmarked for this complex scenario, with a notable exception combines co-trained naive bayes (CoT-NB) BCH coding. In paper, we benchmark the performance of co-regularized least square regression (CoR-LS) semi-supervised multi-class classification. We find it performed consistently significantly better than CoT-NB over eight data sets at different scales. also...
Collective Matrix Factorization (CMF) is a popular model for the joint matrix completion task, but limited by its strong assumption that all matrices share same low rank structure. Recently, promising alternative was proposed with relaxed structures are partially shared. We refer it as Partially (P-CMF). This paper presents first PAC generalization error bound based on P-CMF model. Our contributions tri-facet. First, we derive new single factorization model, which fundamentally improves...